A Deep Dive into Web Performance Testing: Playwright vs. Locust

Welcome to this brains-on workshop exploring two powerful tools for web performance testing: Playwright and Locust. Through guided exercises, participants will simulate user interactions and load scenarios to evaluate the performance of real websites.

---

Workshop Objectives

By the end of this session, you will:

• Understand the differences between browser automation and load testing.
• Use Playwright to measure page load times under various conditions.
• Use Locust to simulate user load and test server-side performance.
• Compare navigation methods and the impact of caching on page performance.
• Visualize and interpret performance results using Python and plotting libraries.

---

Project Structure

.
├── q1a-bts-pw.json           # Playwright Results - Click navigation, no caching
├── q1b-bts-pw.json           # Playwright Results - Direct link navigation, no caching
├── q2a-bts-pw.json           # Playwright Results - Click navigation, with caching
├── q2b-bts-pw.json           # Playwright Results - Direct link navigation, with caching (this and files above are generated after running corresponding scripts)
├── urls.txt                  # Category names and URLs for BooksToScrape
├── q1a-bts-pw.py             # Playwright Script for Q1(a)
├── q1b-bts-pw.py             # Playwright Script for Q1(b)
├── q2a-bts-pw.py             # Playwright Script for Q2(a)
├── q2b-bts-pw.py             # Playwright Script for Q2(b)
├── plots-q1q2.py             # Generation of plots (optional, since qmd file already does that)
├── books-performance.qmd     # Quarto notebook for analysis & visualizations
├── q3-bts-eda.py             # Exploratory Data Analysis
├── q4-bts-loc.py             # Loacust load test script for BooksToScrape website
├── q5-pycon-pw              # Playwright test for measuring load times of PyCon Colombia websites (DO NOT RUN)
├── LICENSE                  # MIT License (NO liability, AS-IS)
└── README                   # Notes and instructions             

---

Requirements

💡 Tip: Install VS Code (for running Playwright and Locust scripts) and Positron (for data analysis).

1. Create and activate your virtual environment

python -m venv venv
source venv/bin/activate

2. Install Python packages

Install all required dependencies using:

pip install -r requirements.txt

Your requirements.txt should contain:

locust
matplotlib
pandas
playwright
seaborn

3. Install Playwright browser binaries

After installing the Playwright package, you must install the browser binaries:

playwright install

This will set up Chromium, Firefox, and WebKit so Playwright can automate them during tests.

---

Exercises

Question 1: Navigation Method

How does the method of navigation (clicking vs. direct links) impact loading times?

Scripts:

• q1a-bts-pw.py
• q1b-bts-pw.py

Question 2: Caching Effects

How does caching affect loading times, with each navigation method?

Scripts:

• q2a-bts-pw.py
• q2b-bts-pw.py

Question 3: Number of Books and Loading Times

Does the number of books in a category affect loading times?

After running Q1 and Q2 scripts:

• q3-bts-eda.py

Question 4: Load Testing

Use Locust to simulate X users accessing the site:

locust -f q4-bts-loc.py --host https://books.toscrape.com -u 25 -r 5 --run-time 30s

Question 5: PyCon Website Load Times (do NOT run!)

Compare how PyCon sites from different years perform:

• q5-pycon-pw.py

---

Analysis & Visualization

Use the provided books-performance.qmd Quarto notebook to:

• Analyze load time distributions
• Run t-tests to assess significance
• Visualize relationships between book count and load time
• Identify slowest categories

---

Author

Rodrigo Silva Ferreira

Feedback & Contributions

Have ideas or suggestions? Please feel free to email me at [email protected] or to open an issue (even if it's just a question!)

---

License

This project is licensed under the MIT License.